Page 1 of 1

Stability Test

Posted: Wed Jul 12, 2017 8:59 am
by ChrisDev
Hello everyone,

I'm currently experimenting with an XL216-Processor to reduce its power consumption.
One way I'm doing this, is to reduce the voltage it's working with.
Is there a known manner to test, if the processor can work stable? I was thinking of writing a short program to do some testing for several days and to check afterwards if any errors occurred.

Greetings,
Chris

Re: Stability Test

Posted: Wed Jul 12, 2017 9:26 am
by plex
I can tell from personal experience that a core voltage bellow the specification of 0.95V will definitely create problems and I have seen here in this forum at least another case where a low voltage was creating intermittent errors.

Re: Stability Test

Posted: Wed Jul 12, 2017 10:43 am
by ChrisDev
Could you specify the kind of problems?
And did you also reduce the system frequency or have you just lowered the voltage, in your personal case?

I'm asking this, because in a paper I've read they've used voltage and frequency scaling in order to achieve a lower power consumption.

Re: Stability Test

Posted: Wed Jul 12, 2017 11:00 am
by plex
I did not do it on purpose. I had a problem with the switching regulator supplying the 1.0V and it was dropping to ~0.93V(with more ripple than it should have as well). The problems that I observed was in USB communications and debug communication which would only some times. The other problem that I read about occurred with errors in the ethernet communications if I remembered correctly.

Re: Stability Test

Posted: Wed Jul 12, 2017 1:30 pm
by infiniteimprobability
here's a link to the other case where low voltage was causing erratic behaviour

http://www.xcore.com/viewtopic.php?f=26&t=5787&p=29389